Re: Debian OCR software that actually works?
On Mon, Mar 08, 2004 at 12:54:29AM -0800, Number Six wrote:
> On Sun, Mar 07, 2004 at 10:26:33PM -0800, Kenward Vaughan wrote:
> > On Sun, Mar 07, 2004 at 06:59:58PM -0800, Number Six wrote:
> > > The output is horrible and unusable -- probably 40% of the text is
> > > misscanned.
> > >
> > it. I believe it is much more sensitive to things like rotated text
> > and font size than you find in commercial ventures. IIRC there were a
> > number of tweaks one could perform in its setup, too.
>
> The other stuff I've read said it is missing a text autostraightener.
> "apt-cache search straighten" shows nothing.
>
> I don't mean to gripe, but OCR accuracy went to >90% everywhere around
> the time the 486 came out. I'm sure GOCR can be made to do equally as
> good as all the rest, but autostraightening text seems like a pretty
> darn important thing.
True. Also easy for us to say, since we are not doing the coding.
Since no one pays 'em for the effort they put into the project, and
they need to put food on the table, things get split up accordingly.
Some absolutely free software is absolutely great. Others work only OK.
The nature of the problem _and_ the people donating their time are
obvious factors going into this. I can only judge the app's usefulness
to me.
> The docs don't help; what other tools did you use?
The docs were the only thing I had then. Other than that... my fingers
on the keyboard. :D
Kenward
--
In a completely rational society, the best of us would aspire to be
_teachers_ and the rest of us would have to settle for something less,
because passing civilization along from one generation to the next
ought to be the highest honor and the highest responsibility anyone
could have. - Lee Iacocca
Reply to: